For the web residencies by Solitude and ZKM on the topic »Engineering Care,« the artist and researcher Pedro Oliveira proposed the project On the Apparently Meaningless Texture of Noise, in which he examines accent recognition software in asylum procedures, and challenges narratives of care around it. »Reimagining care must begin with reimagining institutions. Can we think about the future without addressing matters of care left behind by acts of dehumanization and dehistoricization?« For the web residencies he created a »sound essay,« to serve as an entry point to or a conversation piece on the instrumentalization of voices and accents, on bodies as the site for the production of »evidence.«
Schlosspost: Your project On the Apparently Meaningless Texture of Noise (2019) for the Web Residencies by Solitude and ZKM is a continuation of your work A Series of Gaps Rather Than a Presence (2019), which deals with accent recognition software in asylum procedures. Could you introduce us to your artistic practice as a sound artist, and based on that, elaborate on your inquiries with so-called accent recognition software and explain where did this interest for voice biometry technologies comes from?
Pedro Oliveira: Despite being originally trained as a designer, my work has always been rooted in exploring, studying, and producing sound. It emerges from a deep interest in listening and its relation to articulations of power. I had the honor and luck to do my undergraduate degree in a public university that hosts one of the best design courses in Brazil, with plenty of possibilities for experimentation and discussion of design as a language so my shift toward sound and listening as a way of thinking design (or vice-versa) was more of a natural process that followed me all the way through the completion of my PhD here in Germany in 2017. I do not consider myself a designer, though, and I still have a bit of uneasiness in calling myself a »sound artist« because I think that my work is heavily research-based, albeit conveyed through sonic/artistic means. I say this because I am not interested in sound as sound, but more on the contingent arrangements of sound, meaning, power, affect, and so on. What sound does to bodies and how, and what such doing constructs in the world.
My interest in what I call »sonic biometry« emerges within this idea of what sound does to a body, but also as to how sound becomes this site in which discourses around the production of subjectivities and a so-called »objectivity« often overlap. I make it a point of disconnecting my work from the notion that listening is a purely subjective act, or that sound is more «pure« or »primal.« There is no primacy of sound – there are contingent arrangements (cultural, political, social, material) that inform what it is that sound can do, how, and to whom. In my PhD dissertation I explored these ideas by analyzing the articulation of sound and racialized police violence in Brazil, unweaving the colonial construction of a racialized listening (something that other scholars e.g. Jennifer Stoever in her book The Sonic Color Line have also explored) through designed artifacts (in my case a sound bomb, a jukebox, and an acoustic wall). However, me being a Brazilian researcher working from abroad also entails the idea that I am talking about coloniality as an ontological force that (supposedly) can only happen »there,« whereas the sustainment of coloniality has its origins and concentrates its power »here.« The use of accent recognition software against asylum seekers in Germany, and within this so-called »refugee crisis,« is a clear example of this concentration and articulation of power, so when I stumbled across this information I decided to focus on it and explore it in relation to what I had done before.
»My interest what I call »sonic biometry« emerges within this idea of what sound does to a body, (…). There is no primacy of sound – there are contingent arrangements (cultural, political, social, material) that inform what it is that sound can do, how, and to whom.«
Schlosspost: Through an evaluation of the language profile by an accent recognition software, the human voice becomes evidence, and the accent becomes a placeholder for the detectability of the body. Why do you perceive this software as »acts of dehumanization and dehistoricization«?
PO: The idea that human traits are in any way measurable, classifiable, rankable, and taxonomizable is a violent, colonial construct. It emerges from the notion that a) certain populations and their ways of being are somehow »closer to nature« than others, but also that b) the lives, bodies, objects, languages, and practices of these populations had to be collected, extracted, and »made sense of« by European colonizers. This desire for »understanding« the native denies agency, denies the right to opacity (as Glissant would say), and creates a narrative which is purely told by eyes, ears, and words of these so-called »experts« who had to develop fictional grammar and vocabularies to be able to speak about things they could not (and still cannot) fathom. It is a denial of agency, of the very possibility of agency, and thus a removal from history.
Accent recognition software or any other biometric technology for that matter, in its attempt to essentialize human traits and classify them according to specific mathematical principles under the guise of »objectivity,« is basically a continuation of this process. For instance, any database containing samples of a specific language or accent is bound to be a congealment of a specific articulation of an accent, connected to a specific geopolitical situation and a specific time period. It is not, in any way, an »objective« or »accurate« representation of that accent, let alone of an entire group of human beings. However, this method of identification becomes but a placeholder for identity. Identifying becomes an act of producing a contingent, fictional identity that denies, deconstructs, dismantles any possibility of agency. It only exists insofar as it remains exterior to itself.
Schlosspost: In your project proposal you claim that »reimagining care must begin with reimagining institutions.« You ask yourself, »How do we build ground-up counterpolitics against the institutional infrastructures of control made possible by automated systems under the guise of »humanitarian care«?« In your opinion, what shift should the institutions of »humanitarian care« undertake for a better future?
PO: The very first thing would be to rethink (or to follow my colleague and dear friend Mahmoud Keshavarz, to repoliticize) the idea of »humanitarianism« as it is connected to care. What is the «human« that is subject to »humanitarianism?« Because this entails the creation of a specific category of »human« in need of care – and that new »human« can only be defined by the material conditions created by borders and the violent systems that enforce them. A better future – or any future, for that matter – can only be possible by abolishing borders. So it is a convoluted problem that requires not the reimagination of »humanitarianism« but a reimagination of the infrastructures that create borders in the first place. Once we abolish borders, we also repoliticize and rehistoricize the idea (or the need for) »humanitarian care,« because only so we cease to produce a specific type of humanity that is always defined by the necessity of »being cared for.« These ground-up counterpolitics I mention are already in place, they are happening as we speak.
Schlosspost: Where does your project On the Apparently Meaningless Texture of Noise responds to the (non-)ethics of automated care and the products that manifest the instrumentalization and economization of care?
PO: Artistic work is a form of adding yet another layer of complexity to that conversation, an aesthetic engagement which may produce different forms of counterpolitics (exposing, breaching, occupying spaces, repoliticizing spaces). This project I am undertaking in this residency has two goals: one is to offer a form of counterarchiving to these sounds, and probe the possible relationships of them with text materials (bureaucratic documents, academic papers, poems, etc). The other is to use the space of this project as an attempt to bring out different affects made possible by listening to the timbral matter of the human voice. So this project takes the shape of a short »sound essay« which seeks to be an entry point to or a conversation piece on the instrumentalization of voices and accents, on bodies as the site for the production of »evidence.« But the project itself cannot exist without a dialogue with other types of praxis dealing with similar subjects. In that regard, I reject any claim that it is creating something which is, in fact, already articulated in everyday praxis by those encountering and fighting the violence of the border industry with their own bodies.
»Once we abolish borders, we also repoliticize and rehistoricize the idea (or the need for) »humanitarian care,« because only so we cease to produce a specific type of humanity that is always defined by the necessity of »being cared for.««
Schlosspost: How do you intervene, appropriate, and repurpose the training datasets of such systems and how do you make sure that you are avoiding a violent and dehumanizing use of the material yourself? And what aspects of this particular data, i.e. sonic quality and/or materiality of voice interests you the most?
PO: As a researcher I believe the conversation around ethics comes first, and it is far more important than any aesthetic desire I may have as an artist. First and foremost, I treat the dataset as a rudimentary form of sonic archiving, because, as I explained, it entails the congealment of a given language at a given geopolitical configuration at a given time. Any archive, sonic or otherwise, is bound to be incomplete, and whereas this is a given to most scholars working with archives, within the deployment of these datasets in biometric technologies they become statements of a fabricated »group identity.« So the questions that we need to ask are: how do we listen to these archives? What is at stake when such archives become the site of and for artistic and scholarly intervention? Who is listening? It entails a responsibility for curating a specific moment in which these sounds are, again, occupying a specific space for a specific group of listeners. The dehumanizing character of biometric technology becomes explicit when these voices are reproduced in a space – be it in the training or testing process – in which they are rendered devoid of the possibility of meaning. But they can also reproduce these same violences if presented in an artistic space without attuning our ears to these questions. The focus on the supposedly »inherent« qualities of sound – and how that could translate back to an essentialized categorization of ethnic or geopolitical identity – is a reproduction and repetition of the history of racism and its connection to the development of science.
For me, the question is how to find in the timbral matter of voice openings for different aesthetic interventions, rather than just reproducing them. So focusing on aspects that reinforce the »humanness« of a voice – stops, musings, coughs, breaths, lipsmacks, etc. – is a way for me to approach these voices differently and carefully, and not merely reproducing these violences translated onto a different, artistic space. It is a matter of creating something that, following Saidiya Hartman’s work, does not speak over the archive, but rather uses the gaps in them – literal or otherwise. These gaps are the sites of productive intervention, because they might reveal things that were not there in the first place, or offer an entry point for a longer, much more complex conversation.
Schlosspost: Under what premises do you create your compositions based on ambisonic recording methods? What does your compositional practice look like? Do you work with a particular software and/or modular synthesizers?
PO: My compositional practice is based a lot on long improvisations, false starts, and a good amount of editing; then I usually take these edits and reprocess them in real time using effects, creating thus a »second layer« performance out of them – also mostly improvised but this time more or less structured. Using Ambisonics, however, is rather new to me and it comes as a result of my residency at IASPIS in Stockholm this past summer. While there I was also a guest composer at EMS, where I was introduced to Ambisonics as a compositional tool;  it made me think of sound in a different manner as it is rendered in space. For this specific collection of sounds my main compositional tools were the Buchla 200 from EMS, as well as a little Eurorack system I have at home. I edit in Ableton Live mostly, and because of that I am working with Envelop4Live as the framework for Ambisonics.
»Focusing on aspects that reinforce the »humanness« of a voice – stops, musings, coughs, breaths, lipsmacks, etc. – is a way for me to approach these voices differently and carefully, and not merely reproducing these violences translated onto a different, artistic space.«
Schlosspost: Structurally spoken, the short compositions themselves seem to reproduce a multitude of »languages« and create a choral effect with loose endings. Was it intended that the sounds change between anthropocentric and technological soundscapes?
PO: I would not necessarily make a distinction between »anthropocentric« and »technological« soundscapes, as I understand voice and language to be in themselves technologies, and in that, very powerful ones. In that sense, is it really possible to trace a line between a form of sounding that is manipulated by vocal cords and one done by a synthesizer? That distinction can oftentimes be really blurry not only from a listening standpoint but also from a conceptual one.
Open-ended composition is something that I am really keen on exploring because I believe that the continuity of sound, and specially with this specific type of sounds I am working with, is fundamental to shape the experience of listening. Again thinking about the gaps, it is through a deep engagement with those sonic materials as one listens that diverse meanings, textures, rhythms, and other possibilities emerge. It is meant to be listened to with time – but how long is strongly dependent on the listener and their particular engagement with the sounds that are being presented to them.
Schlosspost: So far you have performed your compositions and sound installations primarily in spatial and performative contexts. How has this shift to the web influenced and/or challenged your work process? How does it feel to »perform« in this different context?
PO: I think a web environment is the perfect place to think of duration. Because how work unfolds in time depends a lot on the conditions in which the it is presented to the person navigating it. It demands a different engagement with listening, fundamentally disconnected from performance, for instance. So it makes me think of how shorter compositions might yield a durational listening experience, without necessarily dictating how short or long that experience should be. Incompleteness is perhaps the driving force here, because it talks back to the archive, to the database, and to the system.
Schlosspost: One last question. What does your desktop or workspace look like?
Schlosspost: Please provide some links, book titles, sound files, videos as further reading related to your artistic practices and research topics to share with our readers.
PO: I am always returning to Gloria Anzaldúa’s Borderlands/La Frontera, Édouard Glissant’s Poetics of Relation, and Saidiya Hartman’s text Venus in Two Acts. I am also indebted to Alexander Weheliye’s writings on Phonographies and Habeas Viscus. More specific to this project, as I mentioned above, Jennifer Stoever’s The Sonic Color Line beautifully articulates the idea of racialized listening, and no conversation on biometry can ignore Simone Browne’s book Dark Matter. Last but never least, the work of journalist Anna Biselli at https://netzpolitik.org/ has been fundamental to my research.
I have published two articles/audio papers on this research so far:
http://spheres-journal.org/das-hatte-nicht-passieren-durfen-re-narrating-border-vocalities-and-machine-listening-calibration/ and https://transmediale.de/content/on-the-endless-infrastructural-reach-of-a-phoneme
The interview was conducted by Denise Helene Sumi
- I wanted to give thanks to sound artist and composer Julia Giertz, who introduced me to Ambisonics at EMS.